For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Mojo function

partition_work

def partition_work(task_id: Int, num_tasks: Int, work: Int, work_block_size: Int) -> IndexList[Int(2)]

Partitions work into blocks distributed across num_tasks tasks.

The work is first grouped into work_block_size-sized blocks for alignment and load/store efficiency, then the blocks are split between tasks as evenly as possible.

Args:

task_id (Int): The index of the task to compute the range for.
num_tasks (Int): The total number of tasks sharing the work.
work (Int): The total amount of work to partition.
work_block_size (Int): The block size used to align the partitioning.

Returns:

IndexList[Int(2)]: A pair (offset, length) describing this task's work range.